I hate when people post hyperpartisan reporting because it makes me do homework. In this case, you made me listen to almost an hour of a three hour podcast with three techbros chatting about techbro crap in techbro ways. You owe me years of life.
Anyway, so the conspicuously missing context here is he's asked if they will let go of the subscription model and go after an ad business model instead and he responds "hopefully not" and clarifies that he thinks the AI differentiator from Google search is that it doesn't feed people ads.
He then transitions into saying that you'd need a super hyperspecialized profile for it to make sense and then maybe it could work but they haven't figured out long term memory well enough for that, which is when he talks about why they'd want to have a browser to build that hyperspecialized profile.
This is my least favorite type of misinfo, too, because he's actually kinda saying what they say he's saying, just out of context. But more importantly, because he says some other shit that is more outrageous, too. For example, when explaining why he thinks the subscription business will grow more than the ad business the way he puts it is that "people see it as hiring someone", so they're more willing to spend, and he ponders "how much do people pay for personal assistants and assistant managers and nannies?" and suggests that they'll provide similar services for cheaper to people who can't afford human help.
Which may not be as clickbaity and I get he finds it positive-on-the-aggregate, but is certainly some cyberpunk dystopia stuff that didn't need the out of context quoting to be a thing.